Packet Loss - Avaya Application Solutions Deployment Manual

Excessive jitter will be experienced as either packet loss, if the jitter exceeds the jitter buffer
size, or as additional delay, if the jitter is less than or equal to the buffer size. Jitter may result in
packet discard creating audible voice-quality problems if the variation is greater than the packet
size used (e.g., 20 ms). Jitter buffers should be sized dynamically to give the best quality. If the
buffer is sized statically, it should generally be sized to twice the largest statistical variance
between packets. However, care needs to be taken in the design of the resizing algorithm of
dynamic buffers in order to avoid adverse effects.
Note:
Dynamic jitter buffering can exacerbate problems in an uncontrolled network.
Note:
Many good tools are commercially available to measure jitter, delay, and packet
loss to help monitor and manage the network.
The network topology can also affect jitter. Multiple paths between endpoints with load
balancing enabled can contribute significant amounts of jitter.
The following Avaya products all have dynamic jitter buffers to minimize delay by automatically
adjusting the jitter buffer size:
Avaya G350 and G700 Media Gateways and G650 Medial Gateways with the TN2302
circuit pack
Avaya IP SoftPhone software
Avaya 4600 Series IP Telephones

Packet loss

Packet loss occurs when packets are sent, but not received (or are received too late to be
processed) at the final destination due to some network problem. Packets discarded by the jitter
buffer of the receiving endpoint can also be considered lost from a user's perspective.
Qualifying problems caused by occasional packet loss are difficult to detect because each
codec has its own packet loss concealment method (PLC). Therefore, it is possible that voice
quality would be better using a compression codec (G.729A), which includes its own PLC,
compared to a full bandwidth G.711 codec without PLC.
The proper treatment of packet loss is dependent on several factors such as the following:
Packet loss is more noticeable for tones (other than DTMF) than for voice. The human ear
is less able to detect packet loss during speech (variable-pitch), than during a tone
(consistent pitch).
Packet loss is more noticeable for short, continuous packet loss than with random packet
loss over time. For example, losing ten contiguous packets is worse than losing ten
packets evenly spaced over an hour time span.
Packet loss may be more noticeable for larger voice payloads per packet than for smaller
ones, because more voice is lost in a larger payload.
Packet loss
Issue 3.4.1 June 2005 207